Discussion

In this prospective cohort study of patients with newly diagnosed pSjD, we observed substantial variability in medication adherence during follow-up. Nearly one quarter of the initial cohort did not attend the planned follow-up visit, and among those reassessed, 40% were categorized as having low adherence according to the CQR-5. Importantly, patients with low and high adherence were comparable in terms of baseline clinical and sociodemographic characteristics. In contrast, adherence was associated with how patients balanced the perceived necessity of their medication against concerns about potential harm, underscoring the relevance of cognitive and perceptual factors in treatment behavior.

The frequency of suboptimal adherence observed in our study is broadly consistent with findings reported in other rheumatic diseases. In rheumatoid arthritis, nonadherence rates ranging from 30 to 80% have been described depending on assessment methods [7]. In systemic lupus erythematosus, approximately one quarter to more than half of patients are considered nonadherent [8, 9]. Similar patterns have been reported in axial spondyloarthritis, where patients’ illness perceptions and medication beliefs have been shown to be associated with adherence [10, 11]. In contrast, evidence on medication adherence in pSjD remains limited. A recent study evaluated hydroxychloroquine blood concentrations in patients with Sjögren disease and reported that a considerable proportion of patients had subtherapeutic levels, indirectly suggesting suboptimal adherence [14]. Unlike that study, which focused on a single medication and used pharmacological monitoring, the present study evaluated self-reported adherence together with medication beliefs, disease activity, Sjögren-specific symptom burden, and broader patient-reported outcomes. Although self-reported adherence measures may overestimate true adherence compared with objective methods [23], our findings suggest that adherence-related challenges are also relevant in patients with newly diagnosed pSjD.

Although the relationship between medication beliefs and adherence has been investigated in other rheumatic diseases, data specific to pSjD are scarce. This distinction is clinically relevant because pSjD differs from other inflammatory rheumatic diseases in terms of treatment goals, the frequent coexistence of sicca symptoms and fatigue, often low systemic disease activity, and heterogeneous use of symptomatic and immunomodulatory therapies. In the present study, adherence status was not clearly associated with ESSDAI, ESSPRI, psychological symptoms, fatigue, functional status, or health-related quality of life, whereas the necessity-concern differential differed between low- and high-adherence groups. Therefore, our findings extend the existing adherence literature by showing that the necessity-concern framework, previously described in other chronic rheumatic diseases, may also be relevant in pSjD. However, because adherence and medication beliefs were assessed at the same follow-up time point, this association should be interpreted as exploratory and non-causal.

Recent studies have shown that adherence-related problems are relevant across rheumatic diseases. Neycheva et al. [12] reported declining compliance with biological therapies over 36 months in rheumatoid arthritis, while Wang et al. [13] showed that health empowerment was associated with medication adherence in gout. Although these studies differ from ours in disease population and adherence assessment, they support the broader view that medication-taking behavior is shaped not only by disease-related variables but also by treatment context and patient-level factors. Our study adds pSjD-specific data by evaluating self-reported adherence together with medication beliefs, the necessity-concern differential, disease activity, and patient-reported outcomes.

We did not observe significant associations between adherence and age, body mass index, symptom duration, sex, educational level, smoking status, or employment. This pattern is consistent with observations from rheumatoid arthritis and lupus cohorts, in which sociodemographic variables alone have generally shown limited explanatory value [7, 8]. These findings further support the notion that demographic characteristics may be less informative than patients’ perceptions and interpretations of their illness and treatment.

Systemic disease activity, as measured by the ESSDAI, remained low and stable in both adherence groups and was not associated with adherence status. Similarly, symptom burden, fatigue, psychological status, functional ability, and health-related quality of life were comparable between groups both at baseline and over time. These findings suggest that, in newly diagnosed pSjD, adherence behavior may not be adequately explained by measurable clinical severity or patient-reported symptom burden. The generally low level of systemic activity in our cohort may also have limited the ability to detect potential relationships between adherence and clinical outcomes.

Beliefs about medicines showed the most consistent association with adherence in this cohort. Although the necessity and concern subscales did not differ significantly when considered separately, the necessity-concern differential was lower among patients with low adherence. This observation is in line with the Necessity-concerns Framework proposed by Horne and colleagues [21], which posits that adherence behavior reflects a balance between perceived personal need for treatment and concerns about potential adverse effects. In exploratory logistic regression, higher necessity-concern differential values were associated with higher odds of being classified as adherent. However, the effect estimate had a wide confidence interval and the model explained only a modest proportion of adherence variability. Therefore, this finding should be interpreted as a hypothesis-generating association rather than as evidence of an independent predictor or causal determinant of adherence.

From a clinical perspective, these findings suggest that evaluation of disease activity alone is insufficient to understand medication-taking behavior. Assessment of patients’ perceived need for treatment and medication-related concerns may help clinicians identify individuals who could benefit from more detailed treatment discussions. However, whether interventions targeting medication beliefs can improve adherence in pSjD requires confirmation in larger longitudinal and interventional studies.

Limitations

This study has several limitations that should be considered when interpreting the findings. First, the sample size was relatively small, particularly after loss to follow-up, which may have limited statistical power and increased the risk of type II error. Second, the predominance of female participants resulted in a limited ability to assess sex-related differences and may restrict generalizability. Third, medication adherence was assessed using a self-report instrument, which may be subject to recall bias and social desirability bias and may overestimate true adherence compared with objective measures.

In addition, detailed information regarding specific medications, number of treatments, treatment modifications, and adverse effects was not systematically collected. Therefore, adherence could not be evaluated in relation to treatment characteristics, which may represent an important contextual factor. Furthermore, the CQR-5 has been validated in rheumatoid arthritis populations, and although widely used in inflammatory rheumatic diseases, its performance in pSjD has not been specifically established.

Another important limitation is that medication adherence and beliefs about medicines were assessed at the same time point. As a result, the observed associations should not be interpreted as indicating a temporal or causal relationship. Similarly, logistic regression analyses were performed in a relatively small sample and explained a limited proportion of variance; therefore, these findings should be considered exploratory. To avoid model overfitting, the analysis was deliberately restricted, and extensive multivariable adjustment was not attempted. The wide confidence interval around the odds ratio indicates statistical imprecision. Accordingly, the regression findings should be regarded as exploratory and hypothesis-generating and require confirmation in larger cohorts with prespecified multivariable models.

Finally, although patients were classified as newly diagnosed, symptom duration indicated that some individuals may have had a longer pre-diagnostic disease course. This heterogeneity may have influenced both clinical presentation and adherence behavior.